def __init__(self, *args, **kwargs): if 'model' not in kwargs: kwargs['model'] = 'resources/model.txt' if 'pattern' not in kwargs: kwargs['pattern'] = 'resources/pattern.txt' super(WapitiPOSTagger, self).__init__() option_dict = {} option_dict['pattern'] = kwargs['pattern'] option_dict['model'] = kwargs['model'] self.model = Model(**option_dict)
def load_crf_pos_model(self, path=None, force=False): if not self._crf_pos_model or force: options = {} if path: options['model'] = path else: options['model'] = os.path.join( library_path, "crf_pos_model.txt") if os.path.exists(options['model']): _crf_pos_model = Model(**options) else: e = IOError() e.errno = 2 e.filename = options['model'] e.strerror = "No such file or directory" raise e self._crf_pos_model = _crf_pos_model return self._crf_pos_model
def __init__(self, stanford_postagger_model=None, wapiti_postagger_model=None, jar_tagger_path=None, jdk_variable_path="C:/Program Files/Java/jdk1.8.0_121/bin/java.exe", tagging_model="wapiti"): import platform if platform.system() == "Windows": self.tagging_model = "stanford" else: self.tagging_model = tagging_model self.dir_path = os.path.dirname(os.path.realpath(__file__)) + "/" if stanford_postagger_model is None: self.stanford_postagger_model = self.dir_path + "resource/postagger/NC_model" else: self.stanford_postagger_model = stanford_postagger_model if jar_tagger_path is None: self.jar_tagger_path = self.dir_path + 'resource/postagger/stanford-postagger.jar' else: self.jar_tagger_path = jar_tagger_path if wapiti_postagger_model is None: self.wapiti_postagger_model = self.dir_path + "resource/postagger/UPC_full_model_wapiti" else: self.wapiti_postagger_model = wapiti_postagger_model if self.tagging_model == "stanford": java_path = jdk_variable_path os.environ['JAVAHOME'] = java_path self.tagger = StanfordPOSTagger(model_filename=self.stanford_postagger_model, path_to_jar=self.jar_tagger_path, encoding='utf-8', java_options='-mx5000m') elif self.tagging_model == "wapiti": from wapiti import Model self.tagger = Model(model=self.wapiti_postagger_model)
def __init__(self, patterns=[], **options): from wapiti import Model self.model = Model(patterns='\n'.join(patterns), **options)
def __init__(self, patterns=[], **options): self.model = Model(patterns='\n'.join(patterns), **options)